SQUALL DETECTION AND HINDCAST

VALIDATIONS

L. Renac, Aktis Hydraulics

SEASTATE CCI 2019

Squalls Identification1. Introduction

1. Squalls definition

2. Background and Objectives

2. Use of Satellite Altimeter data

1. Verification

2. Forward speed and event duration

3. Foot print and Averaging interval

3. Extreme Value Analysis

4. Examples

1. Ghana

2. Brazil

Introduction – Squalls definition

Introduction - Background

• Violent short-lived winds

➢ Significant impact on shipping and port industry• Moored / weathervaning ships

• Crane operation

• Human safety

• Measurements are scarce (squalls are short and local)

• Numerical models struggle to capture them (physics / resolution)

• Satellite data has a global coverage over extended period

➢ Valid solution to fill the gap

➢ Compensates for length of time with spatial coverage

Satellite data - Verification

Compare exceedance with available in situ data and identify

threshold selection for squall isolation

Satellite data - Verification

Compare exceedance with available in situ data and identify

threshold selection for squall isolation

Satellite data - Verification

Use satellite imagery to check the location and time of extreme satellite

data samples

GOES-12 optical satellite data for event on 22-Feb-2009 at 12:00 UTC. The arrow indicates the

altimeter measurement location

Extreme Value Analysis

• Based on the Cumulative Frequency Distribution method

• Assumes homogeneity of the population➢ Need to properly select the data threshold

• Uses the Weibull distribution

• Check sensitivity of results to various parameters:• Event duration

• Estimated Time averaging period

• Data threshold

➢ Confirm that the method is stable

➢ Validate against Guidelines for locations with squalls where they are

available

Example – Ghana

Ta = 10 min

Te = 10 min

Ulim= 9.5 m/s

Example – Brazil

Ta = 12 min

Te = 30 min

Ulim = 13 m/s

Hindcast validation

1. Global Validation

1. Maps

2. Taylor diagrams

2. Point Validation

3. Event Validation

1. Scatterometer Overlay

2. Altimeter Latitude

RMSE between model and Altimeter

Taylor diagram of model vs Altimeter for U10

Point Validation - model vs Altimeter

Scatterometer Overlay Model at 12h, Scat at 12h

Altimeter vs Model, Hm0

Conclusion• Squalls:

• A way of extracting very specific information out of a

satellite database

• Data is suitable for preliminary design

• Limitations

• Future plans include verifying results in squall areas

where measurements are available.

• Global validation

• General statistics about model quality

• Event validation

• Case base verification of model performance

Thank You

Questions?

Satellite data – Forward speed

Use satellite imagery to estimate the forward speed and event duration

GOES-12 Water vapor image on 22-Feb-2009 at 06:00 UTC (left panel)

and 15:00 UTC (right panel)

Satellite data – Time Averaging

Satellite averages over space but for design we need a time average

(Ta) - and we need to know what that time is.

Actual Sampling situation

Satellite data – Time Averaging

Fictional Sampling situation

So we compute Ta = L/V with:

• V=forward speed of the squall

• L=footprint of the Satellite

Conclusion• Important input data:

• Altimeter data

• Averaging time (Ta), from forward speed and foot print

• Event duration (Te), from altimeter data and satellite

imagery

• Selection Threshold

• Data is suitable for preliminary design

• Limitations

• Future plans include verifying results in squall areas

where measurements are available.

Results of the sensitivity analysis in Ghana

Sensitivity Analysis of 100yr 10-minute average

wind speed to estimated Storm Duration and Time

Averaging offshore Ghana, value in [m/s] and

percentage of variation

Results of the satellite imagery checks in Brazil

Top 10 events near SergipeInformation extracted from optical imagery for top

10 events near Sergipe

References for validity of the Satellite data under 25m/s

Calibration and Cross Validation of a Global Wind and Wave Database of Altimeter,

Radiometer, and Scatterometer Measurements

I. R. Young, E. Sanina, and A. V. BabaninDepartment of Infrastructure Engineering,

University of Melbourne, Parkville, Victoria, Australia

https://journals.ametsoc.org/doi/full/10.1175/JTECH-D-16-0145.1

33 years of globally calibrated wave height and wind speed data based on altimeter

observations

Agustinus Ribal & Ian R. Young

Scientific Data 6, Article number: 77 (2019)

https://www.nature.com/articles/s41597-019-0083-9

STD between model and Altimeter

Mean error Model vs Altimeter

Sample density Altimeter

Taylor diagram of model vs buoys for Hm0

Taylor diagram of model vs buoys for Tm02

Scatterometer Overlay Model at 12h, Scat at 13h40

Scatterometer Overlay Model at 15h, Scat at 13h40